The UK has ambitious growth plans for AI with huge amounts of investment promised by business and government to meet these goals, but the issue of ‘sustainable growth’ has been a moot point. The enormous amounts of energy to power this AI growth could result in increased energy prices for all consumers and above all could undermine the UK’s Net Zero ambitions. There is a solution. If we were to locate our AI infrastructure close to areas with abundant renewable energy, for example in Scotland, the negative impacts of this exponential growth could be minimised.
- British ‘Sustainable’ Data Centres
In the UK, data centres are springing up in or close to densely populated areas where there’s already existing pressure on water, electricity and land, and these pressures on vital resources will only increase. It is well known that it is more efficient to transport gigabits than kilowatts over distance and in Norway for example, many data centres have been located close to areas of abundant renewable energy, typically hydro power plants or (multi-hydro) substations, away from metropolitan areas but able to connect to them via fast, high-quality fibre cables. In Scotland, this abundant renewable energy comes from various sources: on- and offshore wind, hydro (large and small-scale) solar and tidal generation. What if the data centres needed for the rapid growth in AI could be located close to these sources of renewable energy like in Norway? These data centre facilities, specialising in AI training workloads, could be large-scale of up to 1GW, medium-scale of 40-50MW, or smaller Edge facilities in more remote regions, connected by sophisticated fibre-optic networks. All of these options could capitalise on Scotland’s renewable energy supply, its land, and transferable oil and gas workforce, all together making it an attractive proposition for locating power-hungry AI infrastructure.
2.. Data sovereignty
Increasingly, while the world economy shrinks its globalisation drive in response to more nationalistic regimes’ protectionist policies, data sovereignty will play a more important role. Countries like the UK could (and should) pivot their data strategies to ensure more data is located on these shores rather than, for example in the US or the Nordics, all to ensure greater (data) security and efficiency and low carbon data transfer.
- Energy
Energy efficiency is key to the successful growth of AI in the UK. The energy demands of high-performance compute such as that seen in today’s Nvidia’s GPU Blackwell architecture or Nvidia superPODs will result in a massive increase in demand for power. Some data centres are providing individual racks of up to 100kW+ to meet this demand, which is in stark contrast to the days of 5-20kW per rack (not that long ago). Scotland can provide the answers to these problems with a concerted effort of investment, foresight and political will from both the devolved and national governments. Their aim must also be to reduce the burden of red tape and bureaucracy to facilitate these large-scale infrastructure projects, which will form the backbone of the UK’s technological drive.
- The future of AI in the UK
The UK is pursuing rapid growth in the AI sector. The UK Government’s own AI Opportunities Action Plan states that “AI adoption could grow the UK economy by an additional £400 billion by 2030.” This plan aims to solidify the UK’s position as a global AI superpower, with forecasts suggesting the AI market could inject nearly $90 billion into the nation’s economy by the end of the decade, providing a significant boost to GDP. In another forecast on growth rate valuations, the UK AI trade body UK AI stated that “projected annual growth rates of 22% to over 32%, with valuations expected to hit between $20 billion and $31 billion by 2030–31, although estimates vary”. Whichever figure it is, exponential growth is all but guaranteed, and the UK is well positioned to capitalise on this AI boom
- Energy
The UK has some of the highest energy prices in Europe, which has a hugely negative impact on business and consumers, and is often cited as one of the main reasons for the cost-of-living crisis. The reasons for these high costs are complex, but it is said that the wholesale gas price is the main driver.
The UK’s path to decouple from gas looks some way off, given the intermittency of renewables and the delays and cost overruns of the large nuclear projects currently under construction. Nuclear Small Module Reactors (SMRs) are also some way off, with Rolls-Royce, the winning bidder of the recent SMR competition by Great British Energy, stating that it will “connect projects to the grid in the mid-2030s” in other words, in quite a long time. In the immediate term, SMR technology is not the panacea to our AI energy supply conundrum some think it is.
5.1 AI energy demands
The scale of projected energy consumption of AI compute in the UK is staggering. AI and related infrastructure consumed around 3.6 TWh of electricity in 2020, but if consumption increases twentyfold as per the government’s target, it could reach 72 TWh by 2030 – representing over one-quarter of the UK’s total electricity consumption. A study by Loughborough University in January this year pointed out that the UK’s net zero ambitions could therefore be severely hampered by this growth.
Above: The NVIDIA GB200 Grace Blackwell Superchip can have a configurable Thermal Design Power of up to 2700 watts, which includes two B200 GPUs and one Grace CPU, and supports a rack-level power consumption of up to 120kW (copyright NVIDIA 2025)
The UK government is aiming to deliver a zero-carbon electricity system by 2030, but the dramatic AI-driven energy demand surge could undermine these climate targets. The energy requirements for AI growth will invariably consume a substantial portion of the UK’s renewable energy capacity too. At the present time though, in Scotland there is plenty of renewable energy capacity as indicated by the Department of Energy Security & Net Zero (DESNZ) and the National Energy Systems Operator (NESO)’s commitment to establish new grid connections (in Scotland). Additional investment may also be coming Scotland’s way via the UK Government’s AI Growth Zone investment plan.
5.2 Grid Constraints for data centres
Data Centre enablement is being hampered by slow grid connections. In October, an article in Data Centre Dynamics stated that “the current queue for grid connections is ‘archaic’”. Since then, the National Grid has looked to address this problem by embarking upon a ‘Great Grid Upgrade’, to make large-scale improvements to grid infrastructure with new transmission lines and distribution networks, but this will take time to roll out and will slow the pace of delivery, which may yet hamper the UK’s AI growth plans.
5.3 Scotland
The grid capacity shows significant regional imbalances, with the north of Scotland more favourable to new grid connections compared to other areas. Scotland also generates more renewable energy than it consumes and it typically exports this surplus to the rest of the UK, then internationally via the UK’s interconnector network to Europe and Scandinavia. This abundance has a downside, it can lead to bottlenecks in the grid (due to old infrastructure), which then leads to curtailment of offshore wind farms – a well-documented problem in the British press. Yet curtailment can be an opportunity. What if battery energy storage systems and data centres were located next to the landing stations of offshore wind farms to utilise this ‘curtailed’ power when curtailment was forecast? The benefit of this could be fewer shut-down turbines, improved grid regulation through stability and lower cost per kW/hr for data centre customers.
- Connectivity
Scotland is well-connected to the rest of the country and to continental Europe and Scandinavia via high-quality subsea cables, such as Tampnet North, and offshore wind farms’ communications cables that run alongside the power cables to onshore substations. There is room for improvement on this front. Scotland is well connected but if it is to become a sustainable AI hub, it must be even better connected. One way of doing this could be with improved onshore cabling and telehousing, situated along railway lines, connecting Edge facilities for example to increase their combined compute power.
- Why is sustainability important?
Sustainability isn’t just about the environment, it is also about the way a business operates and the people and the community it serves. If a business is to scale sustainably, it must do so by building and nurturing a workforce that can adapt to a rapidly changing business landscape. The community must not be overlooked either, their support is key to the success of the AI business, indeed they would likely play a key role in determining whether an AI facility such as a data centre could be located near them. But community support can be strengthened if there are additional benefits passed on, such as with local employment, but also through other means such as waste heat recovery extracted from battery storage systems or data centres. Advanced heat recovery can enable a circular economy, stimulating the growth of industries such as aquaculture, heated greenhouses, or district heat networks directly into residential areas.
Despite that, energy, water, land (and construction) are the main pressures presented by IT infrastructure of this scale, and particularly infrastructure designed and built for AI, consuming large amounts of resources and resulting in huge implications for the local community.
- Why should the UK be a ‘global AI superpower’?
The UK does not just have a well-established AI industry, it is entering an acceleration phase with unprecedented government backing and private investment, positioning it as a major global AI hub.
The reason for this is due to the following:
- Research excellence
- An established tech ecosystem
- Government support
- Industry presence
- Talent and immigration
- Language and time zone
8.1 World-class Universities
The UK has world-class universities like Oxford, Cambridge, Imperial College London, and UCL that are at the forefront of AI research. These institutions have produced groundbreaking work in machine learning, deep learning, and AI applications, creating a strong talent pipeline.
8.2 London
London in particular has developed into a major tech centre, with a mature startup ecosystem, abundant venture capital, and established connections to global markets. The city’s financial services sector has also been an early adopter of AI technologies.
8.3 Government backing
The UK government has made significant investments in AI through initiatives like the Alan Turing Institute, AI research councils, and strategic funding programs. The government has also developed frameworks for AI governance and ethics that help establish the UK as a responsible leader in the field.
8.4 Big tech
Major tech companies have established significant AI operations in the UK. Google has a huge new office centre being developed in London’s King’s Cross and its DeepMind is headquartered in London. Other giants like Microsoft, Amazon, and Meta have substantial AI research teams in the UK. Additionally, homegrown companies like Arm (before its sale) and newer AI startups have contributed to the ecosystem.
8.5 Highly skilled workforce
The UK has attracted top international AI talent through skilled worker visas and research programs, while also developing domestic expertise through specialised degree programs and industry partnerships.
8.6 Geopolitical juxtaposition
Being English-speaking and positioned between the US and Asian markets makes the UK an attractive base for global AI companies looking to serve international markets.
- Conclusion
Can the UK be a global leader in sustainable AI? The answer is more nuanced as demonstrated above but it can be achieved through a collaborative approach between industry, government and communities. Scotland is well-positioned to capitalise on this AI boom, particularly in the immediate term with its renewable energy abundance.
The longer term differs slightly in that there will be less need to locate close to renewables due to the UK’s net-zero grid, but that outcome is a long way off. For now, we must lay the foundations for building out world-leading AI infrastructure proximate to where the source energy is low carbon and low cost.
The risk of the UK failing to meet its net-zero emissions target by 2030 is high if we are to blindly pursue our AI growth plans at any cost. Suppose we locate multiple data centre sites near offshore wind farms’ landing stations and Edge facilities nearer smaller-scale renewables and fast-track their development? In that case, we may be able to address the energy-AI challenge in the next ten years, or until the UK’s grid becomes fully decarbonised. Additionally, this green energy can be low-cost, which can be passed on to the end consumer, for example, with a low rack price per kWh. All of this combined can result in a more sustainable AI future versus an alternative that relies on fossil fuel expansion.
————————————-
About the author:
Ned Collier is a passionate climate activist, challenging the the assumption that the UK’s path of AI growth is sustainable without a shift in strategy.
1 https://aimagazine.com/news/can-uk-cope-with-20-rise-in-data-centres-by-2030
2 https://assets.publishing.service.gov.uk/media/685bd5d8c07c71e5a87097e5/Energy_Trends_June_2025.pdf
3 https://www.world-nuclear-news.org/articles/rolls-royce-smr-named-as-uks-selected-technology
5 AI Energy Council minutes: Monday 30 June 2025 (HTML) – GOV.UK
8 AI Energy Council minutes: Monday 30 June 2025 (HTML) – GOV.UK